| Summary: | 碩士 === 國立成功大學 === 公共衛生研究所 === 98 === Background: Although occurrences of dengue epidemics were associated with mosquitoes, studies assessing the relationship between entomologic indices and risk of dengue epidemics had conflicting results.
Objectives: To clarify the relationship between entomologic indices and risk of dengue epidemics in various geographic levels and determine which entomologic indices could be used for prediction of dengue epidemics in the geographic level of neighborhood.
Methods: We conducted our study with three different methods using data from Cianjhen district, Kaohsiung city from 2006 to 2009. The first method is spatial analysis of linear association between cumulative dengue incidences and mean entomologic indices in the geographic level of neighborhood. The second method is time series analysis, measuring the cross correlation between weekly dengue incidence and entomologic indices measured in the preceding weeks in the geographic level of district. The third method was similar to a retrospective cohort design. This cohort consisted of all the neighborhoods in Cianjhen district. Neighborhoods in which dengue epidemics occurred were defined as case neighborhoods. For any month when case neighborhoods occurred, various entomologic indices including household index (HI), container index (CI), Breteau index (BI) and adult index (AI) measured in the preceding two months were compared between case and non-case neighborhoods to determine whether entomologic indices were related to occurrences of dengue epidemics, in favor of its prediction potential of dengue epidemics. ROC curve analysis was used to find the cut-off value of entomologic index of interest in predicting dengue epidemics.
Results: Linear association was found between various entomologic indices and cumulative dengue incidence only in year 2006 by method 1. In method 2, linear association was found between dengue incidence and the entomologic indices with the time lag of four to six weeks in year 2006, 2008 and 2009. With method 3, neighborhoods with higher BI relating to dengue epidemics next month were mainly seen in three months when dengue epidemics just occurred in the year of interest. Using ROC curve analysis, the cut-off value of BI =9.5 was estimated to have sensitivity of 60% and specificity of 56% in predicting dengue epidemics in the geographic level of neighborhood.
Conclusion: All the current commonly used entomologic indices had limited value in predicting dengue epidemics in the geographic level of neighborhood. More comprehensive vector surveillance would be helpful to explore the value of entomologic indices in predicting where dengue epidemics would occur.
|
|---|
